Skip to main content

Table 2 Performance of WeederH on the ABS promoter set

From: WeederH: an algorithm for finding conserved regulatory motifs and regions in homologous sequences

χ2Score Threshold

%pred

Nc

Sc

Sp

None

98.37

99.70

100.0

0.02

0

78.90

95.77

98.68

0.23

1

51.04

83.91

91.72

0.52

1.5

45.79

81.42

89.07

0.57

2

41.93

78.26

86.75

0.61

2.5

39.53

76.35

85.10

0.64

3

37.55

75.24

84.11

0.66

4

34.55

73.96

82.45

0.69

4.5

33.26

72.42

80.79

0.70

5

32.30

71.56

79.14

0.71

5.5

31.44

70.77

78.48

0.72

6

30.58

68.96

77.48

0.73

8

27.51

64.81

72.18

0.75

FP

37.18

70.32

79.47

0.65

FPSIG

31.55

62.61

70.19

0.71

phastCons

25.97

50.02

55.51

0.73

  1. Performance of WeederH at different χ2 score thresholds on the 99 orthologous promoter sets retrieved from the ABS database. %pred indicates the percentage of nucleotides of the reference sequence belonging to a conserved motif output by the algorithm. Nc indicates the nucleotide coverage, the percentage of nucleotides in annotated sites matched by a predicted motif; Sc indicates the site coverage, the number of annotated sites matched (at least by 75% of their length) by a motif predicted by WeederH; Sp is specificity. The next two rows report the performance of the Footprinter algorithm on the same set, using parsimony score alone (FP), and with the introduction of significance estimation (FPSIG). Finally, the bottom row shows the same data for the phastCons most conserved tracks available at the UCSC genome browser (on genomes: human and mouse, March 2006; rat, November 2004).